Development of Enhanced Lateral Flow test Devices for Point-of-Care Diagnostics

Lateral flow Imunoassays (LFIA) are common, simple to use point-of-care devices for the diagnostic market. Conventionally LFIAs are limited in their complexity since they are optimized for minimally trained operators. Paper-based analytical devices (PAD) are advanced sensors based on a wide range of recently developed techniques for complex analytical methods. In this research, a point-of-care (POC) immunosensor was developed based on techniques adapted from lateral flow and paper-based analytical devices. Alternating layers of paper and tape were used to expand the common 2D design of lateral flow tests to 3D in order to enable complex fluid flow control. Four fluidic valves were integrated for automatic sequential loading of three different fluids to a detection area. Fabrication processes, reagent concentrations, materials and device geometries were optimized and a chip-yield of 92% was achieved. A three step alkaline phosphatase (ALP)-based enzyme-linked immunosorbent assay (ELISA) procedure with Rabitt IgG as model analyte was used to prove the working principle of the sensor. After optimization of crucial assay parameters practicability was verified by visual detection of signal development on nitrocellulose membrane after reaction of ALP and NBT/BCIP with a good detection limit of 4.8 fm.

diagnostics and diagnostic devices, immunoassays and antibody recognition reaction advantages for paper as sensor substrate and lateral flow test devices.

Point of Care Diagnostics and Diagnostic Devices
Point-of-care (POC) devices allow rapid diagnostic tests to be performed at the site of patient care facility. This means the test can be done in the hospital, the emergency room, a physician's office or at home by minimally trained personnel. And, the results are available immediately rather than waiting for hours or days for the results to come back from a central facility [51].
Point of care testing is a fast growing area with a growth rate around 10% in clinical diagnostics which will be one of the biggest driving forces for the future of the in-vitro diagnostics market [54]. According to the market analyst Frost & Sullivan, the US market for Point-of-Care testing devices will increase from a revenue of $2.13 billion USD in 2009 to $3.93 billion USD in 2016 [11]. The market is thereby driven mainly by two high-growth segments, infectious diseases and coagulation monitoring. The infectious disease market is growing due to increasing infections, detection of new diseases and mutations. The coagulation monitoring market is growing dramatically because of expanded testing in patient homes and the growth of patient services [11].
Therefore, diagnostic testing devices have to be lightweight, portable and easy-touse to perform complex biological tests at the site where they are most needed [54].
Microfluidic systems are suitable for those developments since they can be designed to operate from small volumes of complex fluids with efficiency and speed and without the requirement for highly trained personnel [31].
Several companies around the world market rapid tests. They have developed a variety of devices and technologies which reduce the test times to hours or even minutes [51]. Those technologies can be grouped in three different categories: Permanent integrated instruments; pure disposables; and permanent instruments that use disposable components [61].
Permanent integrated instruments are designed for a high-throughput work, with fast and accurate results but even when those devices would be cheap enough, they could not be considered as point of care devices because trained personnel are needed. Carry over between two tests has to be prevented by rinsing the component with cleaning solutions and also frequent calibration is necessary to keep the settings with the standards even when they use microfluidic components [61].
Disposables are analytical tests based on a disposable substrate (e.g. paper) they are normally based on a microfluidic device (e.g. Lateral flow test) and they rely on relatively inexpensive components and reagents which can be produced as commercial off-the-shelf (COTS) products on large-scale production methods, to be relatively affordable. They can be designed for detection of antigens or antibodies and are usable with a wide range of specimens. Most of them are developed to be stable at ambient temperatures without refrigeration for more than a year and the analytical performance for some of them are comparable to reference-level laboratory methods [29].
Disposable tests provide POC diagnostics in areas without access to well-equipped and well-staffed clinical laboratories. Users can quickly learn to perform such disposable-based tests without the requirement to be repeatedly retrained. Because of this, disposable rapid tests are the one diagnostic technology which has been successfully used in the modern military and the developing world. However, complex and expensive approaches do not deliver the needs of the majority of the world's people suffering with infectious diseases, which have access mostly to poorly resourced health care facilities [60].
Disposable tests currently on the market have a number of disadvantages. They are still not as sufficiently sensitive, specific and accurate as laboratory results. They also usually provide just a yes/no answer.
Disposables with a reader are a compromise between disposables and professional instruments. The sample, the process reagents and the waste remain in the disposable while the reader is used to add more complexity to the test by using electrically driven valves and pumps to control the fluid flow of different fluids within the disposable. The advantage over integrated instruments is that the reader does not need to be cleaned between two samples and that calibrates can be stored on the disposable. This compromise allows high performance with low per-test cost [61]. This approach is the perfect POC solution for hospitals where several tests have to be done daily at the patient's bedside. But the higher prices for the readers in comparison to just disposable tests keeps this approach from being widely accepted for at home use.
And, the need for electricity for the pumps and valves is a disadvantage when the test has to be run in developing countries in places without easily available power sources.
For a portable instrument the power could come from automobile generators, photocells, hand-generators, or stored in the disposable [61]. But all these points increase the complexity in comparison to disposable test devices.
Therefore, more complex diagnostic tests based on inexpensive disposables are needed to fulfill the requirements of POC applications at the patient's home and for the developing world.

Justification
One approach for more complex disposable devices with a better sensitivity is to use advanced lateral flow devices which are able to incorporate multiple fluids for the test. With multiple fluids, it is possible to perform more advanced Immunoassay protocols with disposable lateral flow tests. As example, for an ELISA immunoassay assay (vide infra) at least a substrate is needed as another input fluid in addition to the sample [23].
Devices on the market normally consist of a plastic housing with a developing solution pot which contains the second input fluid (e.g. substrate) that is heat-sealed with laminated aluminum film [23]. The user has to rapture the seal in order to release the developing solution. Their major disadvantage is that the user has to rapture the seal at the right time after the sample has been added to the strip test.
Using the paper based fluidic valve technology developed at the University of Rhode Island by Dr. Hong Chen et al. [3] at the microfluidics laboratory of Professor M.
Faghri (vide infra), it is possible to develop lateral flow test devices with more than two fluids that are self-triggered after a certain amount of time. Such advanced lateral flow test strips are capable of conducting ELISA on paper without operator intervention, except for the application of the sample fluid.

Paper Based Analytical Devices
Advanced sensors based on a paper substrate are called paper based analytical devices (PAD) or lab on paper devices (LOP). They have recently gained increasing interest. For decades paper was used for analytical chemistry but lately it was rediscovered as substrate for sensors. This is because paper offers many advantages including biocompatibility, biodegradability, price and availability (see Table 1.1) which makes this material first-choice for development of disposable sensors and integrated sensing platforms [37].

Immunoassays
Immunoassays are a suitable technique for the direct detection of targets of clinical Interest. They rely on the ability of an antibody to recognize and bind to a specific macromolecule in a lock and key mechanism.

Antibodies and Recognition Reaction
Antibodies (AB) are large Y-shaped glycoproteins produced by the B-cells of the immune system to identify and deactivate potentially harmful targets such as viruses or bacteria. They belong to the group of immunoglubins (Ig) and they possess the  [20].
To bind to an antigen the antibody binding-site contains a paratope that is specific for one particular region of 15-22 amino acids on the antigen called the epitope [13]. There are two types of immunoglubin light chains which are called lambda (λ) and kappa (κ) and five types of heavy chains (α, δ, ε, γ, and μ) [20]. The type of heavy chains defines the class of immunoglobulins (IgA, IgD, IgE, IgG, or IgM ) [42].
In human serum approximately 85% of the antibodies belong to the IgG class at a concentration of 8-18 gL -1 , the dimeric IgA (0.9-4.5 gL -1 ) and the pentameric IgM (0.6-2.8 gL -1 ) can also be found [44]. The molecular weights of antibodies are 150 kDa for IgG and IgD, 900 kDa for IgM, 150 or 600 kDa for IgA and 190 kDa for IgE. The chemical composition of the reachable surface of an average antibody is 55% non-polar, 25% polar, and 20% charged [25]. Three groups of antibodies are typically produced: polyclonal, monoclonal and fragments of monoclonal antibodies [44].  part is separated by enzymatic cleavage [44].
The recognition reaction between the paratope, the antigen-binding region (Fab) and the eptitope, the surface structure on the antigen is mainly driven by four different non-covalent binding reactions. Those are electrostatic attraction between corresponding charges, van-der-Waals forces because of electron-density fluctuations, hydrogen bonds between electronegative atoms and hydrophobic interactions between nonpolar carbohydrates [44]. For example the usual number of hydrogen bonds in an Antibody-Antigen complex is acknowledged to be around 10 [21].
The probability of an antibody to bind to a specific antigen is called affinity and it is described by a dissociation constant, KD. KD is the ratio of the association rate constant kassoc and the dissociation rate constants kdiss. For monoclonal antibodies, studies [2 16, 17] have found that the dissociation rate constant has a wider variation than the association rate constant (kassoc = 105-107 M -1 s -1 ). The affinity can also be approximately derived from the law of mass action (Equation 1.1). The affinities for antibodies found in the literature vary between 10 -5 M -1 and 10 -12 M -1 [44].

Lateral Flow Test Strips
This section provides information on the key aspects of the design of lateral flow immunoassay (LFIA) also called lateral flow tests (LFT) with respect to the materials used and their integration with the assay conditions. Figure 1

Membrane
The membrane is the most important material used in a lateral flow test strip. For lateral flow test strips, the membrane must irreversibly bind capture reagents at the test or control lines. Physical and chemical attributes of the membrane affect its capillary flow properties which affects the reagent deposition, assay sensitivity, assay specificity, and test line consistency [34]. The binding characteristics of the membrane are defined by the polymer from which the membrane is made. Commonly used polymers and their binding characteristics are presented in Table 1    When reagent are applied to nitrocellulose membranes, chaotropic agents such as Tween 20, Triton X-100, glycerin, polyvinyl alcohol (PVA), polyvinylpyrrolidone (PVP), and polyethylene glycol (PEG), which might be used to inhibit unspecific binding or reduce the background noise should be minimized or avoided completely until after the capture reagents have been immobilized and fixed. Otherwise these compounds can physically interfere on molecular level between the protein and nitrocellulose and affect the signal development negatively [34].

Sample Pad
The main task of the sample pad is to ensure a uniform distribution and to control the flow rate of the sample to the conjugate pad. According to Millipore [34] the sample pad can be treated with reagents such as proteins, detergents, viscosity enhancers, and buffer salts to perform multiple tasks:  Increase sample viscosity to improve flow properties  Enhance the ability of the sample to solubilize the detector reagent.
 Prevent nonspecific binding of the conjugate and analyte to downstream materials.
 Chemical modification of the sample to ensure immunocomplex formation at the test line Woven meshes and cellulose filters are the two commonly used materials as sample pads. Woven meshes or also called screens have a very low bed volumes, which is why they retain small sample volume, normally around 1 -2μl/cm 2 and they also have good sample distribution properties [34]. Because of this they are used for applications where limited sample volume is available. Besides this meshes are relatively expensive compared to other porous material and the low bed volume is also a disadvantage when the sample pad should be pretreated with different reagents.
Cellulose filters on the other hand are inexpensive and have large bed volumes, which is why they are used when large amount of blocking agents, detector reagents, release agents, pH and ionic strength modifiers or viscosity enhancers have to be loaded to the sample pad. The disadvantage of cellulose filters is the bad contact behavior with different materials because of this a sufficient and consistent contact might have to be ensured by compression with a housing.

Conjugate Pad
The main task of the conjugate pad is to store the dried detection reagents until a liquid test sample is applied to the sample pad and then ensure uniform transfer of the detection reagent and test sample onto the membrane.
According to the membrane manufacturer Millipore [34] the ideal conjugate pad material has to comprise the following attributes.

Absorbent Pad
The absorbent or also called wick or waste pad is used to keep a uniformly capillary flow through the membrane in the right direction and at a proper flow rate. Without or with a too small absorption pad the sample will flow back in the membrane and could raise the background or possibly cause false positives [34]. Absorption pads are commonly fabricated from non-woven, cellulose fiber sheets in variety of thicknesses and densities to suit the needs of the assay

Housing
A housing is not required for accurate assay functionality but there a different reasons why many manufactures choose to place the lateral flow tests into a housing.
The most obvious reason is, to ensure proper operation by forcing the user to apply the sample in the sample pad. For over-the-counter products it also protects the membrane from contamination through splashes. The Housing is also used for labeling to provide important information to the user (e.g. position of test and control line).
Internal pins and bars in the housing are used to keep the strip test in the right place and compresses the materials together to ensure repeatable fluid flow conditions [34].

Objective and Outline of the Thesis
The overall goal of this project is to develop a highly sensitive ELISA based lateral flow test device by using fluidic valves to trigger multiple fluids automatically in a sequential manner. As first step of the project, the knowledge for point-of-care diagnostics and paper based analytical devices is established, following by the development of a multifluid lateral flow test and the integration of an ELISA procedure.
So, this thesis is comprised of six chapters in the following orders: Chapter 4, findings and discussion, presents and discusses the results of this study.
Including different fabrication methods and variation of reagents.
Chapter 5, findings and future work, summarizes all the chapters of the thesis and addresses recommendations for future research.

CHAPTER 2 -LITERATURE REVIEW
This chapter reviews the current status of analytical methods based on Immunoassays including different labeling-detection techniques, designs for immunoassays and validation methods. Also paper as substrate for sensors is discussed and the newest approaches for paper based analytical devices including fabrication techniques and fluid flow manipulation and calculation are being presented.

Analytical Methods Based on Immunoassays
Immunoassays are an important technique for developing highly sensitive sensors.
The targets cover hormones, proteins, metabolites, drugs, tumor products, antigens and antibodies to infectious agents [8]. Polyclonal or monoclonal [42] antibodies are used to detect the target. Current immunoassays have the ability to detect down to 10 -13 mol/l of analyte [8]. Immunoassays have been developed since 1950 and the development still continues [20]. The most recent and commonly used immunoassay technologies and validation methods will be presented in this chapter.

Labeling-Detection Systems
In order to detect or visualize the antibody-antigen complex, which is typically bound to a surface, one of the antibodies needs to be labeled by a marker that can generate a signal. This is mostly the last antibody which is called the detectionantibody. Those detection markers can be of different structure and composition they can be bound covalently or adsorptive and can generate direct (e.g., optical, absorption) or indirect signals (e.g., enzymatic reaction of colored products).
According to Seydack et. al. [44] the ideal detection marker has to meet the following requirements:  Simple and sensitive detection   studies [2] also report the use of fluorescent labels in lateral flow test devices instead of the colored particle labels that lead to a better Limit of Detection (LOD) but require the use of a reader box.

Immunoassay Designs
All Immunoassay protocols can be grouped into direct or indirect (Figure 2.1) and into competitive and not competitive assays (Figure 2.2). The majority of immunoassays are performed with three different general approaches: competitive assay with either immobilized antibody or immobilized antigen approach or noncompetitive two-site (sandwich) assay [50].

Figure 2.1: Indirect (A) and direct (B) immunoassay adapted from [50]
In a direct assay ( This assay architecture is called sandwich assay because the analyte is "sandwiched" between two antibodies. This design is reported to provide a better sensitivity and is the preferred method in the field [20].

Figure 2.2: Competitive immunoassay is based on the competition of two reagents. A) Immobilized antibody approach, B) Decreasing Signal intensity with increasing analyte concentration for competitive assays, C) Immobilized antigen approach adapted from [50]
In a competitive assay the analyte competes with another antigen for the binding to the antibody. This principle leads to a decreasing intensity with increasing analyte concentration (

Figure 2.3: Non-competitive assay A) Sandwich assay with analyte sandwiched between capture and detection antibody B) Increasing signal with increasing analyte concentration for non-competitive assays adapted from [50]
In non-competitive assays the signal increases with increasing analyte concentration ( For most immunoassays the sandwich design is preferred. But when the analyte has a too low molecular weight and can't react with to two antibodies at the same time one of the other presented methods has to be used [8].

Surface Binding Techniques
Immunoassays require one type of antibody or antigen to be immobilized on a solid surface while the other reagents remain in the reaction buffer or sample matrix [44]. Most used materials for the solid phase are nitrocellulose or nylon membranes which are used for test strips and pre-or untreated polymer (as example polystyrene) which is used for microtiter plates [44]. To keep the functionality of the protein consistent, the binding to the solid has to be as adsorptive as possible. Therefore

Figure 2.4: Different alternatives to immobilize antibodies on a solid surface A) Anti-antibody bounding B) Antibody binding proteins C) Streptavidinmodified surface and biotinylated antibody bounding [50]
For the anti-antibody approach less specific polyclonal antibodies, which bind to the Fc part of the desired monoclonal antibody are coated to the surface and the antibodies are bounded together. This approach can't be used in sandwich assays since the polyclonal antibody will react with every monoclonal antibody from the same animal. In a different method antibody binding proteins (e.g. protein A, G, or L) can be immobilized on the surface to hold the desired antibody since these proteins bound to the Fc part (Protein A and G) or the Κ-type light chain (Protein L) of the antibodies.
Sandwich assays are not possible with these techniques.
An indirect method to immobilize antibodies for sandwich assays can be applied by using the biotin-(strept)avidin system. Here the antibody is conjugated to biotin and the surface is coated with avidin or streptavidin. The antibody is then held by the biotin-(strept)avidin reaction. The advantage of this method is the simple and well understood process of labeling the antibodies with biotin which barely influences the recognition properties [14].

Assay Validation
Validating of immunoassay is an important tool during the development of immunoassay applications and it is also a requirement of the European Directive 98/79 EC on in vitro diagnostic tests approved by the European Parliament and Council to bring applications to the market [44]. The following definitions, in accordance to the ICH Guideline "Validation of Analytical Procedures: Text and Methodology Q2(R1)" will provide a background of the validation procedures:

Specificity or selectivity
Specificity is defined as the ability to clearly assess the analyte in the presence of components which could be expected to be present (e.g. impurities, degradants) [47].
Other literature [48] further differentiates between specificity and selectivity: Specificity is an evaluation of the response to a single analyte in contrast to selectivity which is the evaluation of a response to a group of analytes that may not be differentiated from each other.
It should be demonstrated, that the assay results are not affected by typical impurities therefore the pure analyte has to be contaminated with a specified amount of impurities and the test result of the purified and the contaminated analyte have to be compared.

Accuracy or Trueness
The accuracy or trueness of an immunoassay specifies the closeness of agreement between the value which is accepted either as a conventional true value or an accepted reference value and the value found. The accuracy should be determined by using at least nine trials over a minimum of three concentrations in the specified range. The accuracy is described as percent recovery by the assay of known added amount of analyte in the sample [44]. Otherwise the accuracy can be described with the difference between the mean and the accepted true value together with the confidence intervals [47]. If the accuracy is a controversial issue international reference material can be used to prove the accuracy of an assay [44].

Precision
Precision is defined as the closeness of agreement (degree of scatter) between a series of measurements obtained from multiple sampling of the same homogenous sample under the prescribed conditions. The precision should be described as standard deviation, variance or coefficient of variation obtained from a series of measurements [47]. The precision has to be calculated for three different levels: The precision should be determined using a minimum of nine trials over the specified range for the procedure or using minimum of six determinations at the maximum test concentration. For all precision levels the conditions of the trials have to be included to the precision data as specific as possible.

Limit of Detection
The detection limit is the lowest concentration of an analyte in a sample that can be detected by the assay. The three most common approaches to determine the Limit of detection are listed below. Other Approaches than those listed may be acceptable too [47]. The Signal-to-Noise Ratio is determined by comparing the measured signal of samples with a negative control. The detection limit is reached, when, the Signal-to-Noise Ratio falls below a specified limit. A ratio between 3:1 or 2:1 is usually considered acceptable for determining the detection limit [47] l) Standard Deviation of Response and Slope A more accurate and refined method than the methods described above uses the Slope S of the calibration curve and the standard deviation σ of the response. The detection limit is then expressed with following equation:

Limit of Quantification
A quantitative or semi-quantitative assay relies on the opportunity to detect the exact or a range of the analyte amount in a sample in contrast to a qualitative assay which is only capable of detecting whether there is analyte in the sample or not. In order to develop a quantitative assay the Limit of Quantification is an important number for comparison of assays. The Limit of detection is defined as the lowest amount of analyte in a sample that can be quantitatively determined with suitable precision and accuracy [47]. The procedures to detect the LOQ are the same as for the LOD. A visual evaluation is applicable for non-instrument tests such as lateral flow test.
A Signal-to-noise ratio of 10:1 is acceptable for the LOQ and for the method which is using the standard deviation of the calibration curve and the Slope of the response following equation can be used [44]: Also the blank determination applies for the limit of quantification [45]. The LOQ is expressed as the analyte concentration corresponding to the sample blank value including ten times the standard deviation:

Range
The range describes the analyte concentration interval in which the immunoassay has been proven to have a suitable level of precision, accuracy and linearity [47].

Robustness
Robustness is defined as the resistance of an immunoassay against the influenced by variations of the assay parameters (e.g. temperature, pH, humidity). It is an indicator for the reliability during regular usage [47].

Limitations of Lateral Flow Immunoassay
Lateral flow immunoassay are the oldest technique for paper based analytical devices and can be traced back to the 1950s [57]. They were designed as easy to operate rapid diagnostic devices for the point of care market. Paper based analytical devices, can be classified as standard LFIA when they are composed according to section 1.5 and operated without prior sample preparation and without additional steps other than the sample application. Because of their simplicity LFIA have some major disadvantages compared to recent developments advanced paper based analytical devices (vide intra) such as miniaturization of sample volume requirements below microliter level, sensitivity or multiplexing [57]. To compare LFIAs with recent developed paper based analytical devices various performance parameters for LFIAs on the market or recent published in the literature were summarized in Table 2.2. The table list also some more advanced LFIA which are still based on the standard LFIA principle with a conjugate release zone and a reaction membrane but also with some more sophisticated principles to enhance the test results.

Optical Detection: Methods and Detector Systems
Optical detection is the most inexpensive and universal method [22] and it is a perfect application for paper based analytical devices since this substrate offers a bright, high-contrast, and colorless background for the read out of color intensity changes [59] . Several researcher groups [41] also discovered, that paper can give a better sensitivity and quantification than other substrate materials. The complex cellulose structure could also lead to some disadvantages. Researchers discovered, that high background signals (e.g. non-specific binding) or signal non-uniformity (e.g. liquid accumulating on the borders of the detection zone) can be a problem [37]. Chen et. al.
[4] discovered that drying of reagent in an incubator at 37 °C can help to reduce background signal. And other research groups mentioned that treatment with poly(vinyl amine), gelatin, poly(acrylic acid), or poly(ethylene glycol) can help to stabilize color development [37].
A big advantage of optical detection methods is the simplicity of the detector systems. The least expensive detector which doesn't require to buy any further equipment is the naked eye. Detection based on the naked eye might be precise enough for the detection of non-or semi-quantitative assays such as Lateral flow immunoassays. For quantitative assays a reading device is essential. Those reading devices could be simple tools like scanners, digital cameras or phone cameras which are available all over the world and which are easily portable. These detectors are inexpensive and simple to use point of care devices. For application where a higher sensitivity, lower limit of detection or limit of quantification is needed, more specialized detector systems are being developed, including spectrophotometers, fluorimeters and gel documentation systems [2].
A notable device using colorimetric detection is a paper based protein detection system developed by Wang et. al. [53]. The group used bromophenol blue for semiquantitative analysis of bovine serum albumin in artificial urine in a tree-shaped ( Figure 2.5, C) self-calibrating detection system. The design ensures uniform conditions of each assay.

Electrochemical Detection: Methods and Detector Systems
Compared to optical sensors, electrochemical sensors are not affected by dust, light or insoluble compounds [19]. Also different research groups found that using paper instead of solid materials for electrochemical sensors the influence of convection of liquids caused by random motion, vibration, or heating can be reduced [37].
Electrodes can easily be integrated into PADs. For example Hu et. al. [19] presented nanoporous gold electrode arrays on cellulose membranes which were used to develop a cost-effective and environment-friendly paper-based electrochemical gas sensor for the detection of oxide (Figure 2.5, A). Other researchers [43] also discovered that the large surface area of the paper on top of the electrode is able to increase the signal response of the sensor. An interesting PAD approach using electrochemical detection is a device based on potentiometric immunoassay to detect IgE [46]. The PAD is built from nitrocellulose paper sandwiched between two silicone rubber sheets connected to electrodes on both sides (Figure 2.5, B). Electrochemical Flow-Injection with low cost components for the detector system (amplifier, voltage regulator, voltage inverter & batteries) [24]. Also multimeters are being discussed to be the next generation of electrochemical detectors [37]. Liu et. al.
[28] recently developed a paper based analytical device for electrochemical detection of adenosine using a digital multimeter for the readout (Figure 2.7).

Energetic Principles
Most paper based analytical devices are based on 2D or 3D microfluidic circuits using the advantage of capillarity for fluidic manipulations like transportation, sorting, mixing or separation of the needed reagents [26]. Also gravity is sometimes used to enhance the fluid flow [24]. The great benefit of this is, that these devices do not need additional energy sources to cause fluid flow and the analytical test will run on its own after the user has introduced the sample to the system. To accomplish multi-stepanalysis and diagnostic procedures some PADs are based on mechanical manipulators like switches [30] or buttons [17] (vide infra). Also some paper based analytical devices need electric energy sources (e.g. to drive UV lamps or to enable electrochemical readout). Because of this several researchers are working on paper based batteries. For example Thom et al. [49] developed a disposable paper-based galvanic cell battery for diagnostic applications in resource-limited settings. The battery is composed of multiple galvanic cells and can be incorporated directly into a multilayer paper-based microfluidic device.

Analytical Principles
There are several analytical principles in the field, which are used to generate either an optical or electrochemical detectable signal. One main principle to generate a signal is chemical reactions. For example Yu et al. [62] developed a PAD based on chemiluminescence signal generation which is used for simultaneous quantification of glucose and uric acid. Their system is based on generation of hydrogen peroxide through the chemical reaction of glucose and uric acid with oxidase enzymes. Hydrogen peroxide is then used to produce light by reacting with a rhodanine derivative ( Figure   2.6, B).
Another important type of analytical techniques are biological reactions especially those which are based on immunoassays (see section 2.1). Immunoassays have the advantage of high selectivity, rapid detection, and the possibility to analyze complex matrices without pretreatment [37]. Using immunoassays on a paper substrate rather than on a solid surface also leads to a higher surface-to-volume ratio and shorter incubation times (e.g. 10 minutes for paper based ELISA [5] vs. hours on solid surfaces) and possibly better limit of detection. The Whitesides group [5] for example developed a paper based method to replace conventional 96-microzone microtiter plates with paper based ones. Their method requires smaller reagent volumes (e.g 3 μl of sample vs. 70 μl on microtiter plates) and less time (51 min vs. 213 min). Their limit of detection for Rabbit IgG was about 54 fm.
Other biological methods are using different strains of bacteria which are able to produce specific enzymes, which, for example have been used in a paper based colorimetric assays [18]. A different analytical principle is based on the piezoelectric effect discovered in oriented cellulose fibers. It was used to develop paper based strain and vibration sensors [37]. Also temperature dependent reactions are used to develop PADs. For example Nery et al. [37] described a method using thermochromic ink to develop temperature sensors. The sensors consist of a series of pixels of various actuation temperatures. When the actuation temperature of the pixel is reached it turns colorless (Figure 2.6, C).

Figure 2.6: Paper based analytical devices. A) Paper based ELISA [5] B) Chemiluminescence assay [37] C) Temperature sensor [37]
However many platforms are using combinations of those different principles. One example is a paper based electrochemical ELISA that was developed again by the Whitesides group. This analytical device uses the immunoassay technique combined with an enzymatic catalyzed electrochemical reaction. It was used to detect rabbit IgG with a detection limit of 3.9fm [58].

Fabrication Methods for PADs
Fabcrication of paper based analytical devices is fairly simple and inexpensive. The

Three-dimensional PADs
Compared to 2D paper based devices (e.g. dipsticks and lateral flow systems) that are based on lateral movement of fluids across paper strips, 3D paper based microfluidic systems are capable to distribute fluids both vertically and laterally. Using a 3D design it is possible to develop PADs with complex microfluidic paths and the capabilities of those low-cost analytical systems can be expanded significantly. Two different approaches have been developed to build 3D PADs.

Layered paper and tape
Martinez et. al. [33] described a method stacking alternating layers of waterimpermeable double sided tape and paper to create three-dimensional PADs. First they patterned the paper layers with hydrophobic wax in order to define the fluid channel and cut holes into the tape do define the area where the fluid has to flow vertically.
Then they stacked the layers together, the holes in the tape were filled with a paste made from cellulose powder and water in order to ensure a good connection between the layers. The advantage of this approach is the simplicity with which these devices are assembled which makes the prototyping of new designs rapid.

Origami approach
Origami PADs (oPADs) first described by Liu et al. [27] are fabricated on a single sheet of paper (e.g. filter-or chromatography paper) and then assembled into a 3D fluidic architecture by folding and sealing. Prior to the assembly the paper is patterned with hydrophobic wax and heat treated in order to create the desired channels.
Compared to 3D devices fabricated from stacked layers of paper and tape this approach is much simpler in particular for automated mass production. For sealing of those devices the group reported a method using a glossy plastic envelope sealed with an impulse thermal edge laminator [28]. That approach avoids adhesives which can lead to contamination or nonspecific adsorption of reagents or targets [52]. They also reported that it is possible to integrate electrodes by screen printing in those devices ( Figure 2.7).

Control of Fluid Flow in Paper-based Devices
In microfluidic PADs, movement of the various fluids is based on capillary flow, this is why mechanisms and equations developed for conventional microfluidic devices cannot be applied. To accomplish multi-step-analysis and complex diagnostic procedures (e.g. ELISA) the fluid flow in PADs has to be controlled and manipulated (e.g. valves). For designing and modeling, the fluid flow in PADs has to be described with suitable equations.

Single-use Buttons
Martinez et. al. [32] developed a single-use 'on' push-buttons for use in programmable 3D microfluidic paper based devices. A small gap between two layers of paper, which is created by a hole in the tape, is used to separate two channels ( Figure   2.8, C). The gap is closed by pressing the two layers of paper together using mechanical force. The gap will stay closed due to in-elastic deformation of the paper. where reagents and samples must be combined in a timed sequence.

Fluidic Timers
Noh et. al. test has to be read out.

Mechanical Switch
Zhong et. al. [64] developed a mechanical switch which can be integrated into a PAD to stop the fluid flow through a channel until the switch is pulled mechanically.
The switches were produced from rectangular holes in the channels and a paper strip that was patterned with wax leaving out an area with the same width as the fabricated channels. After heating, the paper strip was inserted into the cut-out area of the paper device. Pulling on the paper strip could then be used as a switch that restricts or allows the fluid to flow through the fluidic channel.

Fluidic Valve
Chen et. al.  The principle behind this is a simple reaction of two reagents (Figure 2

Modelling of Fluid Flow in Cellulose Substrates
Fluid flow in cellulose substrates can be distinguished for either flow in dried (paper wet-out) or wetted (fully-wetted flow) materials. Bose cases have two be addressed with different mathematical approaches.

Paper Wet-out
For the simplest case, the one-dimensional fluid flow in a porous cellulose matrix

Fully wetted Flow
A flow in a pre-wetted paper channel of constant width can be described by paper [12]; The permeability of paper can also be assumed to be constant, in the range of 3 x 10 -13 m 2 [12]. Since all constants are known the fluid flow only depends on the geometry. For more accurate calculations the permeability for a specific paper can be developed through iteration.

Abundant vs. Constricted Flow
Zhong et.el [64] also studied the condition for abundant (Figure 2  The total volumetric flow through a paper network of N segments of varying widths in series, during fully wetted flow, follows the same form as Ohm's law for calculating the electric current through a circuit with N resistors in series [12]. By

CHAPTER 3 -METHODOLOGY
This chapter is used to present the methods and experiments used to develop and optimize a fluidic circuit and an immunoassay based on the ELISA procedure for lateral flow. Fabrication methods for the device, different circuit designs, assay development, and housing development are addressed.

Chip Fabrication Method
Fabrication of all chips for this study was done using the layered paper and tape method described in section 2.4 [33]. Briefly, all 3D fluidic circuits were built to contain at least three layers consisting of two layers of filter paper with wax printed fluidic channels held together by one layer of double-sided tape. Holes in the double-sided tape filled with hydrophilic material were used to connect the flow layers (Figure 3.1).
Additional layers of one-sided tape were used to cover the channels to prevent contamination, contact, or evaporation. All chips were fabricated in batches of 4 -6 chips. The channels were printed on filter paper using a solid ink printer (Xerox® ColorQube® 8570) with solid wax ink (Xerox® Genuine Solid Ink Black). All Flow channels were cured for 60 seconds at 120⁰C using a Vacuum Oven (Isotemp® Model 280A, Fisher Scientific). Cut outs were made using a CO2 laser cutter (Epilog® Mini 24) and cutting and printing masks were designed using vector graphic programs such as Inkscape and Corel Draw®. To protect the adhesive sides of the tape during handling and cutting processes tape was covered with wax paper (Parchment Paper, Reynolds®).  Biologically inert and chemically stable. Used to dissolve reagents to a high concentration [16].

Treatment of Paper
Pre-Cut Disks Place Disks Assembly To improve the fabrication process three advanced methods based on the one described have been developed during this study.

Disk Punching
This method uses the same treatment for the disks as described before. However, the method of manually handling the disks was replaced with a semi-automatic tool to place all disks at once. The tool (Figure 3.4) consists of three parts: alignment, disk holding, and punching.

Disk-holding Layers
This Method is based on four additional layers (two layers of tape and two layers of filter paper) to incorporate the disks into the chip. The layers of filter paper are used to fabricate and hold the disks during assembly for multiple disk placement in one assembly step. The additional layers of double-sided tape are used to attach these layers to the flow layers of the chip.
The geometry of hydrophobic disk, surfactant disks, and the hole in the doublesided middle layer was maintained according to the method described above. In contrast to the methods described above each disk is printed and treated separately.
Circles were printed for each disk on layers of filter paper (2.5 mm for surfactant disks and 3 mm for hydrophobic disks) and after heating of the wax treated with Tween 20 or Allyltrichlorosilane. The amount of reagents needed for each disk was optimized using additional experiments (vide infra).

Figure 3.5: Flowchart for valve fabrication using layers to hold and align disks
The minimum amount of Allyltrichlorosilane for each disk needed to generate hydrophobic disks was determined in an parallel study at the laboratory of microfluidics of Professor Faghri by W. Föllscher [10]. It was discovered that each hydrophobic disk has to be treated four times with 2 ul of 4.76 vol.% A3CS in Perflurocompound FC-72 in order to achieve permanent hydrophobicity.
After pretreatment the disks were cut out using a laser cutter while a bridge was kept to hold the disks in the layers. The excised area was made as large as possible in order to guarantee tight contact of the flow layer with the middle tape layer.
Additionally, layers of double-sided tape were placed on-top/underneath of the diskholding layers to attach them to the flow layers. The fabrication procedure is shown in  Cutting of disk holding layers and double-sided tape using laser cutter  Assembly by stacking tape and paper layers

Surfactant Cellulose-Powder
This method is similar to the method described in section 3.2.2 but the number of layers needed for fabrication is reduced. Instead of using disks made of surfactant treated filter paper to fill the holes in the middle layer of the chips. The holes are filled with a treated paste made from cellulose powder. Also the disks holding layers for the hydrophobic part of the valve is replaced. In this method the hydrophobic layer is prepared in the same way retaining the equivalent amount of Allyltrichlorosilane for the disks as before but the disks are not cut out anymore.

Optimization of Surfactant
The amount of surfactant was optimized with the experimental setup shown in  Nitrocellulose (AE100, 12 µm, Whatman®) was chosen as a membrane.
Processing of the materials especially cutting and alignment were optimized during the study. The cutting parameters of the Laser cutter (power, speed and fansupport) were altered for all materials until a straight cut without burnt edges at the highest possible cutting speed was achieved.
Specifications and the manufacturer for cellulose or adhesive materials used during the study are listed below:

3D Lateral Flow Test Strip Development
The 3D multi-fluid lateral flow tests presented in this study were developed by modifying the common strip test design presented in section 1.5. Briefly, the basic structure of a lateral flow test including membrane, sample pad, conjugate pad and absorbent pad was maintained and the lateral flow test was transferred to a 3D PAD by adding additional channels, inlets, and a valve mechanism.

One Valve two Inlets Design
As a base for the design preliminary test results for 3D lateral flow test conducted by J. Cogswell at the laboratory of Prof. Faghri were used [15]. The preliminary design ( Figure 3.8) was based on two inlets and one valve fabricated in three layers. The valve operates according to the previously described mechanism (section 2.5.4).

Two Valves two Inlets Design
Two main disadvantages were found for the design described in section 3.4.1.
Firstly the substrate has to travel back through the already wetted trigger channel, and secondly the substrate has to flow through the conjugate pad. The test is slowed down by the fact that the substrate has to travel in the already wetted channel and enzyme The substrate input pad with the first valve underneath was moved to the side of the test close to the nitrocellulose membrane and a second valve (valve #2) was placed before the nitrocellulose membrane underneath the main channel.

Four Valves three Inlets Design
Wash steps are commonly used in immunoassay procedures to wash away unbound antibodies and increase the signal intensity by lowering the background noise [8]. Because of this several researchers are working on methods to integrate wash steps into lateral flow test devices. For example Fernández-Sánchez et. al. [9] reported a significant improvement of the signal intensity using a wash step in a standard LFIA.
Using a wash step the group was able to lower their limit of detection to 2.4fm. Also preliminary results at professor Faghri's laboratory [15] showed that a wash step can wash away unbound antibodies in cellulose substrates.
Because of this, the design presented in section 3.4.2 was extended to three inlets including the sample a wash and the substrate. Two additional valves were added in order to incorporate the third input into the strip test. The principle for the third input is the same as described earlier. The difference is, that the fluid which vertically flows into the trigger channel of the bottom layer divides into two separate channels. The trigger channel is designed in that manner, that the fluid divides shortly before the first two valves with a longer channel to the third and fourth valve. The length of the trigger channel was determined so the second inlet opens after the sample is consumed and the third inlet opens after the fluid from the second inlet traveled into the waste pad.
Additionally, a sheet of blotting paper was placed underneath the bottom layer using another layer of double sided tape for attachment to extend the absorption area and to stabilize the chip. To prevent bridging of the fluid the conjugate pad was placed underneath the channel instead of above. The length of the main channel was kept as short as possible in order to prevent fluid flow that was too slow.

Chip Optimization
Different trigger channel designs were used during the study (Figure 3.10). The length of the trigger channels were calculated assuming paper wet-out (see section 2.6.1). The parameters for the fluid flow timings were found during the assay development and material parameters were determined using flow experiments (vide infra). The chip design was also optimized with regards to waste material, size minimization, complexity of fabrication, and manufacturing time by changing the geometry of the chip and fabrication batch.

Assay Development
The developed assay was built on the alkaline phosphatase based enzyme linked imunnoabsorbent assay procedure.
The following section describes the procedure shown in Figure 3.11 in detail and addresses the techniques used for preparation and implementation of the assay. Also the optimizations conducted during the study are described. The section can further be used as a manual for repeating the experiments.

Assay Preparation Procedure
Mouse monoclonal (SB62a) and polyclonal antibodies to rabbit IgG labeled with alkaline phosphatase were purchased from abcom® as detection and goat polyclonal and mouse monoclonal (31213) antibody to rabbit IgG were purchased from Pierce® as capture antibody. For blocking of unspecific sites SuperBlock® blocking buffer in TBS was purchased from Thermo Scientific.
In order to enable the highest possible sensitivity, the stock solution of capture antibodies was used for preparation of the detection area. The capture antibodies were  Chip assembly (after drying of material)

Assay Implementation Procedure
Rabbit IgG was purchased from Thermo Scientific and diluted in blocking buffer for the sample solution. Dilutions with antigen concentrations from 1 μg/ml to 1 ng/ml were prepared and dilution factors covered three orders of magnitude (1 X 10 3 ) to reduce the deviation error. The solutions were vortexed in between each dilution step for at least 60 seconds in order to ensure homogeneous distribution of the antigens.
Blocking buffer was used as sample fluid for negative controls.
BCIP®/NBT tablets 2,3 were purchased from SIGMAFAST™ as specific substrate to the alkaline phosphatase label and one tablet of BCIP®/NBT was dissolved in 10 ml ultra-pure water to produce the substrate solution.
The first assay experiments were conducted with the two fluid design described in To document the results of the experiments all tests were recorded using at least 600 dpi scans. The images were taken immediately after finishing of the tests. The signal intensity was measured as mean grey value using the image processing program ImageJ.

Optimization of Conjugate Release
The conjugate release was optimized by changing various parameters, such as sugar concentration in the pad, geometry of the conjugate pad, blocking and materials.
The impact of sugar to the assay was investigated, in a parallel study in this topic and the results showed that a sugar concentration of 20wt.% in the antibody dilution with equal proportions of trehalose and sucrose leads to the optimal antibody release [10].
To achieve optimal antibody release, all conjugate pads for this study were prepared using these previously determined amounts of sugar.
In addition, the conjugate release was optimized by valuating glass fiber material with binders and without binders for the conjugate pad, blocking of conjugate pads before applying antibodies, the amount of fluid needed to release the conjugate and the size of the conjugate pad. In order to compare the release for different conjugate pads, an experiment was designed where the conjugate pad was washed with increasing amounts of fluid and the signal intensity of the release was measured.
Therefore, the tested conjugate pad was first prepared with the antibodies according to section 3.6.1 and placed onto blotting paper. Afterwards the conjugate pad was washed several times with steps in the range of 20-40 μl of fluid. After each washing step the conjugate pad was moved to a different place on the blotting paper. Next, 10 μl of substrate was applied to each place where the conjugate pad was washed and the signal intensity of the developed signal was observed.

Optimization of Detection Antibody
The detection antibody has an important impact on assay sensitivity and background noise. To optimize this parameter the concentration of the anti-rabbit IgG antibody in the conjugate pad was varied between 10 μg/ml to 80 μg/ml while all other assay parameters were kept constant. The assay was run using triplicate 500 ng/ml samples per detection antibody concentration. To further optimize the signal polyclonal and monoclonal antibodies were evaluated as detection antibodies. To compare the results the signal-to-noise ratio was determined by reading out the mean grey value of the scans using ImageJ.

Optimization of Capture Antibody
In order to enable the highest possible sensitivity the stock solution of capture antibodies were used for preparation of the detection area. To further optimize the signal polyclonal and monoclonal antibodies were evaluated as capture antibodies.
Triplicate sets with at least five different analyte concentrations for both kinds of antibodies were conducted. The results were examined for the quality of signal development and the nonspecific binding on the detection spot.

Dose Response
After optimization of the immunoassay, the performance of the system was tested by observing of the dose response and by estimating the limit of detection and the limit of quantification. To do this, the signal response for triplicates of 8 different concentrations ranging from 1 ng/ml to 5 μg/ml was measured using the methods described earlier. In order to develop a POC device a housing was needed to store the substrate and wash solutions and according to section 1.5.5 a housing is also necessary to compress different materials together and ensure consistent assay conditions. Experimental results (vide infra) also showed that compression of material can increase the valve performance particularly for the valve fabrication method described in section 3.2.2. A basic housing design with compression for the valves and the connection between the materials was developed by Föllscher [10]. Then, the bottom layer of the housing with the inserted chip is placed onto the upper layer sealing the inlets by pressure. The wax paper strip has to be pulled to activate the immunoassay sensor after which the sample can be added to start the test.

CHAPTER 4 -FINDINGS AND DISCUSSION
This chapter presents the knowledge gained during the development of a paper based analytical devices, which uses fluidic valves to trigger multiple fluid flows in order to autonomously conduct ELISA procedure. Different fabrication methods are compared and the results of the immunoassay optimization are addressed.

Material Processing
Proper cutting parameters are very important for the fabrication process and the lateral flow test itself. Cutting power that is too high or too low can lead to burned edges of the material. Particles resulting from the burned material can be dissolved from the fluid flow and be transferred into the channels or detection area leading to discoloration. The various materials used in this study have a different tendency to burn, for example, a relatively low cutting power (3%) is needed to cut nitrocellulose membranes because of the high affinity to burn, whereas slightly different cutting parameters (+2% cutting power) can cause nitrocellulose to burn. Also the cutting speed is an important parameter for fabrication. Since many masks (channels, absorption, membrane, tapes etc.) have to be cut and one cutting process can take up to 15 minutes (e.g. cutting of disks holding layers) rapid cutting is preferable. On the other hand, the accuracy decreases with increased cutting speed resulting in rough edges. Fan support can help to prevent burning of material but it can also ruin the cutting process and make it inaccurate due to shavings and particles that have been dislodged. In cases where those parts cannot be secured with tape or weighed down, the fan support should be deactivated. For some materials, especially tapes or glass fibers with binders, cutting can result in smoke development. In those cases cutting without fan support is not possible. The parameters presented in Table 4.1 are the results of the optimization process. Cutting using the listed parameters results in sharp unburned edges at the highest possible speed without smoke development.

Development and Optimization of the Fluidic Circuit
Optimization for the fluidic circuit was done with regards to reliability and repeatability of the fluid flow, particularly with reference to the valve opening performance. Different fabrication methods were investigated for their impact to the valve performance as well as different trigger channel designs. Also the amount of surfactant was optimized to improve the system reliability and repeatability.

Comparison of Fabrication Methods
The methods for the production of valves presented in section 3.2 have different advantages and disadvantages (Table 4.2) which will be discussed in the following.
Assembling a chip with four valves using the initial method by manually placing each disk for the valves one after another takes about 50 minutes for a batch of 6 chips. The accuracy of placing the disks is highly dependent on the experience of the constructor and a consistent alignment over different chips is not possible. These circumstances result in high deviations in the valve opening and therefore in high valve failure rates (vide infra). Also the manner of pretreating a whole paper with the reagents could lead to variations of reagent concentration in the material itself resulting in poor valve opening behavior. However, the design and preparation of the fluidic circuit design and valves is quick, easy and very little waste is produced. Another benefit is that the valve disks can be mass produced for future use. it can only be used for one particular chip design. If the chip has to be changed the tool has to be redeveloped and refabricated as well. Also it has been found, that the fabrication with rapid prototyping is not satisfactory with respect to tolerance accuracy.
Although more layers are needed for the fabrication of valves with the "diskholding" method and the design for printing and cutting masks are more complicated, it still offers the opportunity to place several valve disks in one assembly step without the need of a specific tool. Once a chip design for this method is established, changes to this design are easily possible. Since this methodology requires treatment of each individual disk, it offers the opportunity to control the amount of reagents very precisely. This approach was used to optimize the exact amount per disks needed for optimal valve opening behavior (vide infra).
The major disadvantage of the "disk-holding" method is the relatively large gap between the layers caused by the extra tape layers needed to connect the disk holding layers to the channels. Because of these gaps no repeatable valve opening behavior could be reached (vide infra). But in another study [10] it was found, that compression of the layers using a housing can help to improve valve performance. The "disk-

Impact of Fabrication on Reliability and Repeatability
Reliability and repeatability are important factors for analytical devices, particularly for the market admission of medical products [47].  It was also observed, that the method of fabrication for the valves has an impact on the valve opening performance. This is probably due to the transfer of surfactant to the hydrophobic disk, which is dependent on the amount of surfactant, the material, and the contact between surfactant and hydrophobic disks.  As presented in Figure 4.   (close to the saturation point of the cellulose material; >170 μg per disks), could lead to diffusion of the surfactant. It should also be noted, that valves in chips assembled with such high concentrations of surfactant load failed after storage that lasted longer than three days. Due to this, approximately 150 μg of surfactant per disk can be considered as an optimal amount that will allow consistent valve opening, but also allows for extended storage.

Optimization of Trigger-channel Design
As discussed above, the biggest impact on the chip-yield, and therefore the assay reliability, is the probability of bad valves in a chip. Only one bad valve results in failure of the entire chip. Therefore, a decreasing chip yield with an increasing number of valves was found (cf.  be more likely to fail than valves that are placed in a separate channel.

Determination of Chip Geometry and Materials
The required trigger channel length has been determined to allow for the correct valve timing and to reduce waste. In order to improve the fabrication time, the optimal chip geometries were determined. Finally the developed fluidic circuit was tested with water containing food coloring.

Trigger Channel Length
To calculate the trigger channel length assuming paper-wet out (see section 2.6.1) the effective pore diameter of the used material has to be known. The effective pore diameter for the materials used in the device are presented in Table 4.3. They were derived by the conducted flow experiments and calculated according to section 2.6.  (Figure 4.6) which still matches very well with the calculated lengths.

Chip Geometry and Absorption Area
In order to reduce waste of materials the height of the chip was fitted to the width of the double-sided tape (35 mm) and the width of the chips was optimized to fit as many chips as possible to one sheet of filter paper (8x10 in). The batch size was included in the consideration of the chip geometry in order to reduce fabrication time.
It was found that one layer for a batch of 6 chips (2x3) with the geometry shown below ( Figure 4.5) could be fabricated 8 times with one sheet on filter paper producing as little waste as possible.

Figure 4.5: Description and geometries for a batch of 6 chips after fabrication
Since the maximum chip height was determined by the width of the double sided tape the unused space not taken up by the chip was used for the absorption area. The tape between the various flow layers was cut out in the same geometry as the waste pad and filled with cellulose material to increase the waste pad area. In order to keep a constant fluid flow, fluid congestion at the waste pad has to be avoided. Therefore rapid absorption from the main channel is essential and glass fiber was chosen as material to fill up the gaps in the tape due to of the largest available pore size, which results in the fastest absorption rate (Table 4.3).
Four layers of glass fiber are used to fill the gaps in the tape this results in a total area of 646 mm 2 . The area of the filter paper for the chip, including the main channel, trigger channel, and the inlets is about 610 mm 2 . For both materials together the retention volume for the chip can be approximated to 245 μl.
Since this retention volume is not enough to hold the volume of 310 μl used during the test, the waste pad is increased by a blotting paper underneath the chip (total area of 760 mm 2 ). This has two advantages, firstly the blotting paper provides for stability of the chips and secondly blotting paper offers particularly high retention volumes (see Table 4.3). The blotting paper increases the retention volume for the entire chip to 725 μl (+196%), allowing the sample to be captured and also preventing backflow of fluid into the main channel.

Geometry and Proof of Concept
The pictures taken during a test with water colored with food coloring after 5, 300, 310 and 600 seconds (

Optimization of Assay Parameters
This section presents the results of the assay development and optimization process. The overall goal was to achieve a particularly low limit of detection. Therefore the conjugate release, the detection antibody concentration, and the kind of capture antibody have been improved.

Optimization of Conjugate Release
Influence of sample pad blocking to conjugate release Blocking the conjugate pads prior to adding the detection antibody solution decreases the amount of fluid needed to wash out the conjugate. As can be seen from   This is probably due to non-specific interactions between the cellulose fibers and the antibodies. In larger conjugate pads more surface area is available for the same amount of antibodies resulting in a higher probability of non-specific interactions causing more antibodies to stick to the fibers. But it should be noted that it is not possible to reduce the size of the conjugate pad too much, as smaller conjugate pads are more difficult to handle and align and the contact to other materials can be problematic. An insufficient contact between the channel and the conjugate pad can result in a slow fluid flow or prevent the conjugate release. Because of this, 4x5 mm conjugate pads were assumed to be optimal for the developed test.

Conjugate Material
The comparison of glass fiber material with and without binder (

Polyclonal vs. Monoclonal Antibodies
The use of polyclonal antibodies for the detection zone and the conjugate pad lead to higher signal intensities for lower analyte concentrations than the monoclonal antibodies (cf.

Optimization of Detection Antibody Amount
It was found that the amount of detection antibody in the conjugate pad has an important impact to the signal quality and therefore also determines the limit of detection. As can be seen from Figure 4.11, 320 ng is the optimal amount of detection antibody per conjugate pad in order to get the best possible signal to noise ratio. Too little of an amount of detection antibody leads to decreasing signal quality because of decreasing probability of antibody -antigen binding reactions and therefore decreasing signal intensity. Too high amounts of detection antibodies lead to higher background signals without contributing to the actual signal intensity.

Assay Results and Dose Response
The results for the response of the developed immunoassay to different analyte concentrations are shown below:  The lowest analyte concentration that that could still be discerned from human observation was reached with 6 ng/ml. The next lower concentration that was carried out was 2.5 ng/ml and no visible signal was achieved. It can be seen that the mean intensity of the background varies. This is due to the fact that the fluid flow is not uncompromisingly even, especially for the last fluid, the substrate. Since the substrate contributes in equal proportions to the signal and the background intensity, it is possible to smooth the results by calculating out the background signal. This was done by computing the signal to background intensity ratio.
By plotting the signal to noise ratio, a curve was generated (

CHAPTER 5 -CONCLUSION AND FUTURE WORK
The research demonstrated that automatic sequential loading of multiple fluids to a detection area with enhanced lateral flow test devices was achieved. Using enzymelinked immunosorbent assay (ELISA) and Rabbit IgG as model analyte it was proved, that complex diagnostic procedures can be proceeded autonomously in lateral flow. A prototype of a low cost, time efficient and easy-to-operate point-of-care device based on the developed test was achieved by storing the necessary reagents and the device into a housing.
First, different designs and fabrication methods for paper based devices with fluidic valves were developed and their influence to the valve performance explored.
Fabrication processes, reagent concentrations, materials and device geometries were optimized and the valve opening deviation was reduced to 10 s. Also a Chip-Yield of 92% for devices with four valves was achieved.
Second, the developed methods were used to incorporate a three step ALP-based enzyme-linked immunosorbent assay procedure with Rabitt IgG as model analyte into a lateral flow test. Four fluidic valves were used to control the sequential loading of sample, wash and substrate to the detection area. The feasibility was verified by visual detection of signal development on nitrocellulose membrane after reaction of ALP with NBT/BCIP. Immunoassay parameters were optimized including conjugate release, amount of reagents, detection and capture antibodies.
Through optimization a proof-of-concept device with a good limit of detection at 4.8 fm was achieved. The results may not show as promising as expected since the sensitivity is still limited in the range of common gold-nanoparticle based lateral flow test devices (see section 2.2). But compared to other available methods based on ELISA for example microtiterplates which can reach down to 4 fm [58], this system offers resemblance in the detection limit with simplified operation.

Recommendations for Future Work
This section is used to propose several ideas to further improve the developed system such as signal enhancement, signal amplification multiplexing, fabrication methods or electrochemical detection.

Signal Enhancement
As first step the signal development due to background noise, signal deviation and non-specific binding should be addressed. Further research should be performed for different materials, blocking reagents and blocking duration. One promising material that should be investigated is Fusion 5 TM produced by Whatman [56] the material, based on a single layer matrix technology, was developed to perform all the functions of a lateral flow strip on a single substrate. The manufacturer advertises that this material has outstanding non-specific binding properties resulting in conjugate release of >94%. Therefore this material could be used to replace the conjugate pad and the flow channels to attempt lower background noise and the loss of analyte and antibodies in the conjugate pad or channels due to nonspecific binding. According to Whatman their materials also acts as a membrane for striping antibodies as test and control lines. The antibodies have to be conjugated to latex beads, in order to allow binding of the antibodies to the membrane. With this method the upper layer of the developed multifluid lateral flow test could be completely fabricated from one material reducing fabrication time and fluid flow deviation due to insufficient alignment.

Signal Amplification
To improve the sensitivity and to lower the limit of detection gold nanoparticles (GNP) could be used to amplify signal development. In the current Immunoassay structure the signal development is limited by the restraint to label only one alkaline phosphatase (ALP) enzyme to the detection antibody ( Figure 5.1, A). Several researchers are working on amplified systems for ELISA based immunoassays. For example Munge et. al. [36] are using massively labeled superparamagnetic particles to improve the sensitivity for their detection system based on Horseradish peroxidase (HRP)-electrochemical ELISA. A similar setup to their detection system could be used to improve the developed ELISA based optical immunosensor. Commonly used nanometer sized gold particles could be labeled to the detection antibody and several ALP-enzymes could be conjugated to the particle using the biotin-(strept)avidin system.  Martinez et. al. [32] reported, that 3D Paper based analytical devices, such as the one developed during this study, are perfect platforms for simple integration of multiplexing. Multiplexing for diagnostic devices encounters increasing interest in the literature since those devices can be used to detect multiple analytes simultaneously in one sample or generate calibration curves for the assays with one test [32].
Integrating multiplexing into the developed platform can easily be done by expending the device with additional layers on-top of the current layers to distribute the reagents.

Origami Fabrication Method
Although the fabrication for paper based devices with fluidic valves was optimized and simplified during this study it still takes hours to fabricate a huge amount of prototypes. Liu et. al. [27] developed a method for fabrication of paper based analytical devices which they call origami approach and which does not require double-sided tape to hold the various layers together. They fabricate their devices on only one layer of paper, fold it and laminate it in order to obtain a 3D fluidic device (see section 2.4.1).
This fabrication method could be adapted for paper based devices with fluidic valves.
Disk layers for the valves could be fabricated with the approach presented in section 3.2.2 with the distinction that the disks are not cut out anymore. With this method fabrication could be limited to four layers of paper for channels and disks leading to further optimization of the fabrication time and decrease of the probability of alignment defects.

Electrochemical Detection
By printing electrodes on paper electrochemical detection could be integrated to enable quantification in a simple to operate paper based analytical device. For fabrication of the electrodes screen-printing with conductive carbon ink as reported by Liu et. al. [28] Dr. Constantine Anagnostopouloscould be used. The paper substrate should be nitrocellulose because of the superior biding properties compared to other materials [34]. This also would have the advantage that the developed lateral flow test does not has to be modified to integrate the electrodes since the test is already optimized for nitrocellulose membrane in the detection area. Therefore the electrodes should have a design similar to the one presented in Figure 5.3. The geometry is chosen in that way to simply replace the standard membrane of the test with the one printed with a common electrode for immunoassay applications. The electrode and the wiring is screen printed on the same substrate to have a simple as possible fabrication process. Areas in which fluid flow has to be prevented are covered with hydrophobic wax.